- Email: [email protected]
Global distribution of N2O emissions from aquatic systems: natural emissions and anthropogenic effects
Chemosphere ± Global Change Science 2 (2000) 267±279
Global distribution of N2O emissions from aquatic systems: natural emissions and anthropogenic eects Sybil P. Seitzinger a,*, Carolien Kroeze b, Renee V. Styles a a
Institute of Marine and Coastal Sciences, Rutgers University, 71 Dudley Road, New Brunswick, NJ 08901, USA b Institute of Environment and Climate Research, Wageningen University, Wageningen, Netherlands Received 9 July 1999; accepted 12 October 1999
Abstract Context Abstract: Atmospheric concentrations of nitrous oxide, a greenhouse gas, are increasing due to human activities. Our analysis suggests that a third of global anthropogenic N2 O emission is from aquatic sources (rivers, estuaries, continental shelves) and the terrestrial sources comprise the remainder. Over 80% of aquatic anthropogenic N2 O emissions are from the Northern Hemisphere mid-latitudes consistent with the geographic distribution of N fertilizer use, human population and atmospheric N deposition. These N inputs to land have increased aquatic as well as terrestrial anthropogenic N2 O emissions because a substantial portion enters aquatic systems and results in increased N2 O production. Thus, wise management of N in the terrestrial environment could help reduce/control both aquatic and terrestrial N2 O emissions. Main Abstract: The global distribution of N2 O emissions from rivers, estuaries, continental shelves, and oceans are compared to each other, and to terrestrial emissions, using existing gridded inventories. Rivers, estuaries and continental shelves (1.9 Tg N yÿ1 ) account for about 35% of total aquatic N2 O emissions; oceanic emissions comprise the remainder. Oceanic N2 O emissions are approximately equally distributed between the Northern and Southern Hemispheres; however, over 90% of emissions from estuaries and rivers are in the Northern Hemisphere. N2 O emissions from rivers, estuaries, and continental shelves combined equal oceanic emissions in both the 20°±45°N and 45°±66°N latitudinal zones. Over 90% of river and estuary emissions are considered anthropogenic (1.2 Tg N yÿ1 ); only 25% of continental shelf emissions are considered anthropogenic (0.1 Tg N yÿ1 ); oceanic emissions are considered natural. Overall, approximately one third of both aquatic and of terrestrial emissions are anthropogenic. Natural terrestrial emissions are highest in tropical latitudes while natural aquatic emissions are relatively evenly distributed among latitudinal zones. Over half of both the anthropogenic terrestrial and aquatic emissions occur between 20° and 66°N. Anthropogenic N inputs to the terrestrial environment drive anthropogenic N2 O emissions from both land and aquatic ecosystems, because a substantial portion of the anthropogenic N applied to watersheds enters rivers, estuaries and continental shelves. Ó 2000 Elsevier Science Ltd. All rights reserved. Keywords: Nitrous oxide; Aquatic; Rivers; Estuaries; Continental shelves; Anthropogenic; Global emissions; Geographic distribution
1. Introduction *
Corresponding author. Tel.: +1-732-932-6555; fax: +1-732932-8578. E-mail addresses: [email protected] (S.P. Seitzinger), [email protected] (C. Kroeze), [email protected] (R.V. Styles).
Total global N2 O emissions include a wide range of sources from terrestrial and aquatic systems, with both natural and anthropogenic components (e.g., Khalil and Rasmussen, 1992; Matthews, 1994; Bouwman et al., 1995; Bange et al., 1996). A clear understanding of the
1465-9972/00/$ - see front matter Ó 2000 Elsevier Science Ltd. All rights reserved. PII: S 1 4 6 5 - 9 9 7 2 ( 0 0 ) 0 0 0 1 5 - 5
268
S.P. Seitzinger et al. / Chemosphere ± Global Change Science 2 (2000) 267±279
magnitude of the various natural and anthropogenic sources and their geographic distribution is necessary to quantify factors contributing to the current increases in atmospheric N2 O concentrations, to predict future potential increases, and to develop realistic and costeective strategies to reduce future emissions. There have been a number of analyses of the magnitude and global distribution of terrestrial and deep ocean sources (e.g., Nevison, 1994; Bouwman et al., 1995); however only recently have such analyses been done for freshwater and coastal marine systems (Seitzinger and Kroeze, 1998). N2 O emissions from freshwater and coastal marine ecosystems are estimated to amount to 1.9 Tg N yÿ1 (Seitzinger and Kroeze, 1998), which is similar in magnitude to estimated N2 O emissions from a number of other sources, including arable land, grasslands, and animal excreta (Bouwman et al., 1995). Much of the N2 O emission from these aquatic systems is anthropogenic and thus contributes to the current increase in atmospheric N2 O. In addition, large increases in future N2 O emissions from these aquatic systems are predicted as a result of increases in human activities (Kroeze and Seitzinger, 1998). Bouwman et al. (1995) presented an analysis of the total and latitudinal distribution of most of the known terrestrial sources and Nevison et al. (1995) has done so with oceanic sources. We now extend that analysis by adding freshwater and coastal marine N2 O emissions. In the current paper we compare: (1) N2 O emissions from rivers, estuaries and continental shelves with N2 O emissions from oceanic regions, (2) anthropogenic and natural emissions from aquatic systems, and (3) aquatic and terrestrial N2 O emissions, both natural and anthropogenic. These comparisons are done at a global scale as well as on a latitudinal basis to provide insight into the geographic distribution of the dierent sources. Gridded (1° ´ 1°) estimates of N2 O emissions from published studies are used in these analyses. 2. Methods We calculated N2 O emissions for the following latitudinal zones for the northern and southern hemisphere: 0°±20°, 20°±45°, 45°±66° and 66°±90°. Emissions from oceans: We obtained the gridded (1° ´ 1°) global N2 O emission inventory for ocean emissions from Nevison (1994) through the Emission Database for Global Atmospheric Research (EDGAR) (Olivier et al., 1996). The EDGAR database had been converted from the original 2.8° ´ 2.8° grid of Nevison (1994) to a 1° ´ 1° grid, which resulted in a loss of 0.2 Tg N yÿ1 from the original 3.8 Tg N yÿ1 estimate (Bouwman et al., 1995). That database (1° ´ 1°) is based on NevisonÕs ``best'' estimate emissions which she cal-
culated using the water±air ¯ux calculation of Erickson (1993). It is also available from the Global Emissions Inventory Activity (GEIA) web site (http://blueskies.sprl.umich.edu/geia). We excluded from Nevison's database the N2 O emissions from the area that is considered continental shelf by Seitzinger and Kroeze (1998), i.e., the gridcells with an average depth <200 m. The totals for N2 O emissions from oceans by latitudinal zones then were calculated by summing the N2 O emissions from all gridcells with an average depth >200 m. Emissions from continental shelves: We used the gridded (1° ´ 1°) global N2 O inventory from Seitzinger and Kroeze (1998). The totals for latitudinal zones were calculated by summing the emissions from the continental shelf gridcells within these zones. Emissions from estuaries: We used the gridded (1° ´ 1°) global N2 O inventory from Seitzinger and Kroeze (1998). The totals for latitudinal zones were calculated by summing the emissions from the estuarine gridcells within these zones. Emissions from rivers: We used the gridded (1° ´ 1°) global N2 O inventory from Seitzinger and Kroeze (1998). The total emissions for latitudinal zones were calculated by summing the emissions from the watershed gridcells within a latitudinal zone. This diers from the approach used by Seitzinger and Kroeze (1998) to assign river N2 O emissions to latitudinal zones; they assigned the N2 O emissions from an entire watershed to the latitude of the river mouth for that watershed. Terrestrial emissions: We used the gridded (1° ´ 1°) global N2 O emission inventory for terrestrial emissions from dierent source categories from Bouwman et al. (1993, 1995) which we obtained through the EDGAR (Olivier et al., 1996). It is also available through the GEIA website (http://blueskies.sprl.umich.edu/geia). The terrestrial emission source categories distinguished by Bouwman et al. (1995) include natural soils, arable land, animal excreta, biomass burning, agricultural waste burning, post-clearing eects of deforestation, industry, energy and fuelwood combustion (Bouwman et al., 1995). In the EDGAR database several of these sources have been broken down into subcategories, totalling 16 sources of N2 O. Emissions from a 17th subcategory, fossil fuel combustion from air transportation, were not available as a gridded database from EDGAR, so they were not included in our terrestrial totals. N2 O emissions from this source are 0.003 Tg N yÿ1 , which is equivalent to 0.09% of the total anthropogenic terrestrial emissions (Olivier et al., 1996). We used the aquatic inventories identi®ed above in the current analysis because as far as we are aware they are the only gridded inventories available for rivers, estuaries, continental shelves or oceans. Details of the methods used to estimate emissions from each of these sources and uncertainties in each inventory are discussed in the original publications (Nevison et al., 1995; Kroeze
S.P. Seitzinger et al. / Chemosphere ± Global Change Science 2 (2000) 267±279
and Seitzinger, 1998; Seitzinger and Kroeze, 1998). There have been a number of estimates of terrestrial N2 O emission from particular sources (e.g., Matthews, 1994; Nevison, 1994) and some additional sources or modi®cations to the above estimates (discussed below) since Bouwman et al. (1995). We used the inventories of Bouwman et al. (1995) because of their relative completeness and availability in gridded format. The reader is referred to Bouwman et al. (1993, 1995) for discussion of the uncertainties in those terrestrial inventories.
3. Results and discussion 3.1. Aquatic N2 O emissions N2 O emissions from rivers, estuaries and continental shelves: Global watersheds receive inputs of nitrogen (N) from a variety of anthropogenic sources (Fig. 1). The three anthropogenic sources of new ®xed N are synthetic fertilizer, combustion of fossil fuels and biological N2 ®xation associated with cultivated crops. Together, these sources contribute an amount of N (about 90±130 Tg N yÿ1 ) similar to the natural biological ®xation of N2 in non-agricultural soils (140 Tg N yÿ1 ) (Galloway et al., 1995). A considerable amount of the anthropogenic N inputs to terrestrial ecosystems eventually enters rivers through leaching, runo and direct discharge. A portion of this N in rivers is transported to coastal marine eco-
Fig. 1. Schematic of N inputs to aquatic ecosystems. N inputs to rivers and estuaries are linked to N inputs to the surrounding watersheds from both natural and anthropogenic sources. Continental shelves receive N inputs from both terrestrial sources (river/estuarine export and atmospheric deposition) and from oceanic (e.g., onwelling, advection) sources. Subsequent cycling of N through nitri®cation and denitri®cation pathways results in N2 O emissions from these aquatic systems.
269
systems (estuaries and continental shelves). The biological cycling of natural and anthropogenic nitrogen through these aquatic ecosystems results in emissions of N2 O during both nitri®cation and denitri®cation. The global geographic distribution of dissolved inorganic N (DIN) export by world rivers to coastal marine ecosystems and associated N2 O emissions in rivers and coastal marine ecosystems were estimated in Seitzinger and Kroeze (1998). The detailed methodology is presented in Seitzinger and Kroeze (1998). A brief description of the approach is presented here. DIN input to and export by rivers was modeled as a function of N inputs to the watersheds of rivers from synthetic fertilizer use, atmospheric NOy deposition, human sewage, and water runo. (Input and export refer to ¯uxes.) (N input from the watershed areas below the mouth of a river, for example from cities located on estuaries below river mouths, also was included in the estimated river export.) N2 O emissions in rivers and their estuaries were estimated as a function of denitri®cation and nitri®cation using the emission factors described below; denitri®cation and nitri®cation were estimated as a function of external N inputs to the rivers or estuaries. Emission factors of N2 O for nitri®cation and denitri®cation rates were 0.3% for rivers and estuarine gridcells, except in gridcells with external N inputs exceeding 10 kg N haÿ1 yÿ1 ; for these, an emission factor of 3% was used. These emissions factors are based on a variety of studies which suggest that the ratio of N2 O:N2 ¯uxes is generally within the range of 0.1±0.5%, with ratios up to 6% in heavily polluted sediments (e.g., Nishio et al., 1983; Jensen et al., 1984; Seitzinger, 1988). N2 O emissions in continental shelf regions also were estimated as a function of nitri®cation and denitri®cation, however with a lesser amount of detail than for rivers and estuaries (Seitzinger and Kroeze, 1998). N2 O production associated with denitri®cation was estimated by extrapolating previously estimated denitri®cation rates for three latitude zones (0°±20°, 20°±45° and 45°± 90°) in the North Atlantic (Seitzinger and Giblin, 1996) to shelf areas in those latitudes in other ocean basins. N2 O production associated with pelagic nitri®cation was estimated by applying the depth average rate of nitri®cation measured in shelf waters of the western US to the global shelf area. Continental shelves were de®ned as areas with water depths less than 200 m using the USDOC/NOAA ETOPO-5 bathymetry database (Edwards, 1986; Pilson and Seitzinger, 1996). An emission factor of 0.3% was applied for N2 O for nitri®cation and denitri®cation rates in continental shelf regions. Relative magnitude of all aquatic sources: Model predicted N2 O emissions from global rivers (1.1 Tg N yÿ1 ), estuaries (0.2 Tg N yÿ1 ) and continental shelves (0.6 Tg N yÿ1 ) total 1.9 Tg N2 O-N yÿ1 (Seitzinger and Kroeze, 1998) (Fig. 2). This substantially increases the only other gridded inventory of global N2 O emissions
270
S.P. Seitzinger et al. / Chemosphere ± Global Change Science 2 (2000) 267±279 Rivers 1.1
Estuaries 0.2 Continental Shelves 0.6
Ocean 3.5
Fig. 2. Comparison of global emissions of N2 O from all aquatic ecosystems (no global emissions estimates for lakes). Units: Tg N yÿ1 (emissions estimates from Seitzinger and Kroeze, 1998 and Nevison et al., 1995).
from aquatic ecosystems which is that from oceanic regions (3.5 Tg N yÿ1 ; NevisonÕs (1994) inventory minus emissions from continental shelf gridcells). Global emissions from lakes have not been estimated. Overall the N2 O emissions from freshwater and coastal systems (1.9 Tg N yÿ1 ) account for about 35% of the total aquatic emissions (5.4 Tg N yÿ1 ) (Fig. 2). The uncertainties in the freshwater and coastal marine N2 O emissions estimates are considerable. For rivers, estuaries and continental shelves (1.9 Tg N yÿ1 ), the estimated uncertainty range is 0.9±9.0 Tg N yÿ1 , based on uncertainties in emission factors only (Seitzinger and Kroeze, 1998). The emission estimates were derived by using dierent emission factors, which depended on the external N inputs (0.3 or 3% of nitri®cation and denitri®cation rates; see above). A sensitivity analysis indicated that a twofold increase or decrease in the threshold N input value used to determine the emission factor would not markedly alter the overall conclusions of the study by Seitzinger and Kroeze (1998), given the other model uncertainties. Their analysis furthermore indicated that the calculated emissions from rivers are less sensitive than emissions from estuaries to changes in the threshold N input. Unfortunately, there are few rivers or estuaries where N2 O emissions have been measured in conjunction with N input over the temporal (yearly) and spatial (entire river or estuary) scales necessary to compare with the model output. A comparison of the modeled N2 O emissions from the Amazon River and Tamar and Narragansett estuaries indicated that the modeled emissions agree within a factor of 2±5. However, additional comparisons across a range of rivers and estuaries are needed to validate the modeled N2 O ¯uxes. The estimates for aquatic emissions of nitrous oxide are likely to be more uncertain than emissions from terrestrial systems, primarily because there are many more studies of N2 O emissions in terrestrial systems. Van Aardenne et al. (1998) analyzed the
uncertainties in agricultural emissions of nitrous oxide from terrestrial systems for the Netherlands, following the methodology in Mosier et al. (1998). Van Aardenne et al. (1998) also estimated aquatic N2 O emissions associated with external N inputs (leaching) to aquatic systems from agricultural sources using the methodology in Mosier et al. (1998) which is similar to that of Seitzinger and Kroeze (1998). Van Aardenne et al. (1998) concluded from a Monte Carlo analysis, that of the dierent model parameters analyzed, relatively large contributions to the total uncertainty were due to uncertainties in the emission factors from aquatic systems and agricultural ®elds, and in the estimated fraction of agricultural nitrogen subject to leaching and runo. The latitudinal distribution of aquatic sources: There are obvious patterns in the latitudinal distribution of N2 O emissions from aquatic sources (Fig. 3; Table 1). Oceanic N2 O emissions are approximately equally distributed between the northern (42%) and southern (58%) hemispheres. Approximately 70% of the total continental shelf emissions are in the northern hemisphere. However, N2 O emissions from estuaries and rivers are almost exclusively in the northern hemisphere (96% and 98%, respectively). A more detailed look at the distribution of emissions (Tg N yÿ1 ) by latitudinal zones (0°±20°, 20°±45°, 45°±66° and 66°±90° in Northern and Southern Hemispheres) indicates further dierences in the pattern among aquatic sources. For example, the total amount of N2 O emitted from the oceans in each of the ®ve latitudinal zones between 45°S and 66°N is similar, with total oceanic emissions between 45°S and 66°S being at least twice as great (Fig. 3). Continental shelf emissions also are similar in each of the ®ve latitudinal zones between 45°S and 66°N, although there is a slight trend of decreasing emissions going from north to south. In contrast, N2 O emissions from rivers or estuaries show considerable variation among latitudinal zones with the 20°±45°N and 45°±66°N zones having the highest total emissions. N2 O emissions from rivers, estuaries, and continental shelves combined equal oceanic emissions in both the 20°±45°N and 45°±66°N latitudinal zones. These dierences in latitudinal distributions among aquatic sources are related to a variety of factors. Oceanic N2 O ¯uxes were calculated by Nevison et al. (1995) by coupling sea±air gas transfer coecient ®elds to a map of surface water N2 O supersaturations obtained from cruise data. In general, the latitudinal distribution of surface N2 O re¯ects both wind velocities and the amount of mixing between N2 O rich deep water and surface waters (Nevison et al., 1995). An overall view of the latitudinal distribution of ocean N2 O emission per unit area (¯uxes) was obtained by combining the ocean area (Fig. 4(A)) with the ocean N2 O emission for each latitudinal zone (Fig. 3). A pattern of decreasing N2 O
S.P. Seitzinger et al. / Chemosphere ± Global Change Science 2 (2000) 267±279
271
Fig. 3. Comparison of the latitudinal distribution of total N2 O emissions from rivers, estuaries, continental shelves and oceans Units: Tg N yÿ1 .
emission per unit area of ocean is apparent between 66°N and 45°S (Fig. 4(C)). In the Southern Hemisphere between 45°±66°S there is a large increase in rate per unit area (¯ux) due to upwelling of N2 O rich deep water (Nevison, 1994; Nevison et al., 1995). There is a strong pattern of decreasing continental shelf area going from the Northern to the Southern Hemisphere (Fig. 4(B)); however, the N2 O ¯ux, as de®ned in the current calculations, does not vary greatly among latitudinal zones. This results in the general decrease in total N2 O emitted from continental shelf regions per latitudinal zone going from north to south (Fig. 3). The latitudinal distribution of river and estuary N2 O emissions (Fig. 3) is consistent with the distribution of N inputs to watersheds from anthropogenic activities such as fertilizer use (FAO, 1990 modi®ed by Bouwman et al., 1995), human population density (Lerner et al., 1988 updated to 1990 by Bouwman) and atmospheric NOy deposition (Dentener and Crutzen, 1994) which are dominant in the northern hemisphere between 20°N and 66°N (Fig. 5(A)±(C)). Relative magnitude of natural and anthropogenic aquatic N2 O emissions: In the current analysis we attribute N2 O emissions to natural or anthropogenic sources depending on the origin (natural or anthropogenic) of the N source to the aquatic system that is nitri®ed or denitri®ed. N inputs to oceanic regions are considered to be essentially all natural in that they are primarily from N2 -®xation in the oceans; thus N2 O emissions in the oceans linked to nitri®cation and denitri®cation of that N are considered natural (Table 2). Two potential additional sources of N to oceanic regions are atmospheric deposition and rivers. All N inputs from rivers are likely removed by denitri®cation and burial in
the sediments of estuaries and continental shelves (Christensen et al., 1987; Seitzinger and Giblin, 1996) and thus do not contribute to N inputs to oceanic regions. Anthropogenic NOy deposition to coastal and oceanic regions is estimated at 9 Tg N yÿ1 (Galloway et al., 1995). However, given that anthropogenic NOy deposition is relatively small compared to N inputs from oceanic N2 -®xation (wide range of estimates but generally >80 Tg N yÿ1 ; Pilson, 1998), and that only a portion of the deposition is to oceanic regions, in this analysis it is considered negligible. N inputs to rivers are from a combination of both natural and anthropogenic sources. Knowing what component of the N in rivers is from natural or anthropogenic sources is dicult to estimate because of a paucity of measurements of N in rivers with no anthropogenic activity in their watersheds. In a recent model of DIN export by world rivers, roughly 25% (5 Tg N yÿ1 ) of a total of 20.8 Tg yÿ1 was estimated to be natural N (Seitzinger and Kroeze, 1998). This estimate was made by simply assuming that natural export of DIN is 40 kg N (kmÿ2 watershed) yÿ1 , as estimated by Meybeck (1982) based on available data for relatively unpolluted rivers globally. The actual value is likely to vary geographically (Howarth et al., 1996), however, at present a better estimate is not available. Globally, N2 O emissions from rivers are estimated to be 1.1 Tg N yÿ1 (Seitzinger and Kroeze, 1998). The emission factor used by Seitzinger and Kroeze (1998) to relate N2 O emissions to denitri®cation in rivers varies as a function of N inputs to rivers (see above). Therefore, to estimate natural N2 O emissions from rivers, we assumed a natural watershed DIN yield of 40 kg N (kmÿ2 watershed) yÿ1 (Meybeck, 1982) and applied the approach of Seitzinger
a
278 193 0 17.8 431 920
0.053 0.001 0 0.003 28.1 28.2 0.624 28.7 149
Total anthropogenic terrestriala
Total terrestrial
Total terrestrial and aquatic
4151
2599
1409
560 435 35.5 46.9 1190 2267
31.0 107 194 332
579 157 145 671 1552
45°N±20°N
Total anthropogenic terrestrial includes all terrestrial sources except natural soils.
1956
1195
764
2.91 43.2 229 275
354 113 54.8 239 761
66°N±45°N
0.003 0.109 0.455 0.567
16.0 93.6 1.29 8.83 120
90°N±66°N
3330
2588
534
80.0 181 222 18.4 2054 2555
16.7 5.30 10.7 32.7
525 90.2 13.0 114 742
20°N±0°
N2 O emissions by latitudinal zones (Gg N yÿ1 )
Terrestrial Abiogenic Biofuel Energy/transport Insustrial processes Total abiogenic Biogenic Arable land Animal excretion Biomass burning Agricultural waste burning Natural soils Total biogenic
Aquatic Ocean Cont. shelf Estuaries Rivers Total aquatic
Source
3060
2613
296
25.4 88.6 162 5.34 2317 2598
7.37 1.5 5.4 14.3
342 83.3 7.15 14.3 447
0°±20°S
1299
767
199
18.6 118 32.7 5.02 568 742
2.21 5.94 16.9 25.0
452 75.1 1.39 3.57 532
20°S±45°S
1225
13.1
6.51
0.111 6.33 0 0.006 6.62 13.1
0.012 0.026 0.029 0.067
1193 19.2 0.003 0.362 1212
45°S±66°S
43.7
0
0
0 0 0 0 0 0
0 0 0 0
32.2 11.5 0 0 43.7
66°S±90°S
15213
9804
3210
963 1021 452 93.4 6594 9123
60.3 163 457 680
3493 643 222 1051 5409
Total
Table 1 N2 O emissions summed by latitudinal zones for all sources (Gg N yrÿ1 ). Emissions inventories for oceans from Nevison et al. (1995), other aquatic sources from Seitzinger and Kroeze (1998), and for terrestrial sources from Bouwman et al. (1995)
272 S.P. Seitzinger et al. / Chemosphere ± Global Change Science 2 (2000) 267±279
S.P. Seitzinger et al. / Chemosphere ± Global Change Science 2 (2000) 267±279
273
Fig. 4. Comparison of the latitudinal distribution of: (A) ocean area (106 km2 ); (B) continental shelf area (106 km2 ); (C) N2 O emissions for oceans and continental shelves per unit area of ocean or shelf (kg N kmÿ2 yÿ1 ).
and Kroeze (1998). Natural N2 O emissions from rivers are estimated to be approximately 5% (0.05 Tg N yÿ1 ) (Table 2). The remaining 95% of global N2 O emissions from rivers, 1.05 Tg N yÿ1 , are considered anthropogenic. In addition to DIN, organic N (both dissolved and particulate) is transported by, and transformed within, rivers; however, N2 O emissions associated with transformations of that organic N have not been estimated. N inputs to estuaries were estimated based on modeled N export by rivers. Natural N2 O emissions from estuaries (0.02 Tg N yÿ1 ; Table 2) were calculated using the above estimate of natural DIN export by rivers (5 Tg N yÿ1 ) and the approach of Seitzinger and Kroeze (1998). Anthropogenic N2 O emissions (0.20 Tg N yÿ1 ) were calculated as the dierence between the total es-
tuarine N2 O emissions (0.22 Tg N yÿ1 ) and natural emissions. Therefore, anthropogenic emissions account for 91% and natural emissions only 9% of the N2 O from estuaries. Continental shelf N2 O emissions are a combination of anthropogenic and natural. Again, we estimated the relative amount of natural and anthropogenic N2 O emissions based on the N sources to the shelf. Continental shelves receive total N (TN) inputs from both terrestrial sources and from the oceans (Fig. 6). The oceanic component (e.g., onwelling, across shelf transport) is considered to be natural as discussed above. Terrestrial N inputs to the shelf include river export (direct discharge of large rivers to shelves as well as export of N from estuaries) and atmospheric deposition directly to shelf waters. The relative magnitude of ter-
274
S.P. Seitzinger et al. / Chemosphere ± Global Change Science 2 (2000) 267±279
Fig. 5. The latitudinal distribution (for exoreic watershed areas only) of: (A) fertilizer use (Tg N yÿ1 ); (B) atmospheric NOy deposition (Tg N yÿ1 ); (C) human population density (billions).
restrial and ocean N inputs to shelves globally is not well known. However, an analysis was recently done for the North Atlantic (Seitzinger and Giblin, 1996). For the North Atlantic continental shelf, ocean sources (10±13.5 Tg N yÿ1 ) are estimated to contribute about 1.5 times as much N as terrestrial (rivers+atmosphere) sources (8.2±10.6 Tg N yÿ1 ) (Fig. 6). As noted above, ocean sources are considered to be natural N. The anthropogenic and natural components of river and atmospheric N inputs to the shelf are estimated as follows. Twenty-®ve percent of river DIN export to coastal systems globally are estimated to be natural (see above). We use this percentage to estimate the portion of total river N inputs (6.4±8.8 Tg N yÿ1 ) to the North Atlantic continental shelf that are natural (1.6±2.2 Tg N yÿ1 ) and
anthropogenic (4.8±6.6 Tg N yÿ1 ) (Fig. 6). Atmospheric N deposition (NOy, NHx, dissolved organic N) also contains both natural and anthropogenic components. Pre-industrial (i.e., natural) and current anthropogenic NOy deposition to the global oceans (3.3 and 9 Tg N yÿ1 , respectively) and to continents (4.5 and 22 Tg N yÿ1 , respectively) have been estimated by Galloway et al. (1995), however, the relative contribution of natural and anthropogenic NOy deposition to shelves was not estimated. Since continental shelves are located between the terrestrial and the oceanic regions, we assume that the relative amounts of natural (22%) and anthropogenic (78%) NOy deposition to the North Atlantic shelf are the average of that to continents (17% natural and 83% anthropogenic) and oceans (27% natural and 73%
S.P. Seitzinger et al. / Chemosphere ± Global Change Science 2 (2000) 267±279
275
Table 2 Estimated N2 O emissions (Tg N yÿ1 ) from aquatic ecosystems, including distribution of natural and anthropogenic emission (see text for assumptions) Source
Total N2 O emission
Natural N2 O emission
Anthropogenic N2 O emission
Continental shelves Estuaries Rivers Subtotal Oceans
0.6a 0.22a 1.1a 1.9 3.5b
0.4 (66%) 0.02 (9%) 0.05 (5%) 0.47 (25%) 3.5 (100%)
0.2 (34%) 0.20 (91%) 1.05 (95%) 1.42 (75%) 0 (0%)
5.4
4.0(74%)
1.4 (26%)
Total a b
Seitzinger and Kroeze (1998). Nevison et al. (1995) after omitting emissions from continental shelf gridcells.
Fig. 6. N inputs to North Atlantic continental shelf from estuarine export plus large river discharge, atmospheric deposition, and ocean inputs (e.g., onwelling, advection) (from Seitzinger and Giblin, 1996). The range of inputs from rivers/estuarine export is due to the range in estimates of N removal by denitri®cation within estuaries before N is exported to the shelf (Nixon et al., 1996). Ocean onwelling was calculated by dierence to balance N losses through burial and denitri®cation (from Seitzinger and Giblin, 1996).
anthropogenic). We also, somewhat simplistically, assumed the same distribution for NHx. Of the total NOy and NHx deposition to the North Atlantic continental shelf (1.8 Tg N yÿ1 ; Nixon et al., 1996), we estimate that 0.4 Tg N yÿ1 are natural and 1.4 Tg N yÿ1 are anthropogenic (Fig. 6). (No estimate of organic N deposition to continental shelf areas was available.) Considering ocean, atmospheric and river inputs of N to the North Atlantic shelf, and the distribution of natural and anthropogenic components in each, we estimate that about 66% of the N inputs to the shelf are from natural sources and 34% from anthropogenic sources. Following the approach for rivers, estuaries and oceans, we would then proportion the estimated total N2 O
emissions from this continental shelf region between natural and anthropogenic sources according to the proportion of natural and anthropogenic N inputs. For shelf areas outside the North Atlantic, no comparable analyses of the relative contributions of the sources (ocean, atmospheric, rivers) exist and they are likely to dier in other regions. The oceanic N contribution to continental shelves outside of the North Atlantic is especially uncertain. For lack of additional information, we have used the North Atlantic scenario to estimate the proportion of the total global N2 O emissions from continental shelves (0.6 Tg N yÿ1 ) that is from natural (66%; 0.4 Tg N yÿ1 ) and anthropogenic (34%; 0.2 Tg N yÿ1 ) N sources (Table 2).
276
S.P. Seitzinger et al. / Chemosphere ± Global Change Science 2 (2000) 267±279
Based on the above analyses, anthropogenic emissions dominate the freshwater and coastal marine N2 O emissions with approximately 75% attributed to anthropogenic N2 O emission and 25% to natural (Table 2). However, when ocean emissions are added the pattern reverses; now approximately 26% (1.4 Tg N yÿ1 ) of the total aquatic emissions are attributed to an anthropogenic N source and 74% (4 Tg N yÿ1 ) to a natural N source. 3.2. Comparison of aquatic and terrestrial N2 O emissions Overall comparison of natural and anthropogenic emissions: Bouwman et al. (1995) summarized the global source distribution of N2 O emissions from terrestrial sources and from the ocean. We now compare the updated aquatic N2 O emissions to the terrestrial N2 O emissions. Aquatic emissions account for about a third (5.4 Tg N yÿ1 ) and total terrestrial emissions about twothirds of the total estimated global N2 O emissions (15.2 Tg N yÿ1 ) (Fig. 7; Table 1). The total terrestrial emissions include both biogenic (natural soils and agriculture) (9.1 Tg N yÿ1 ) and abiogenic (biofuel, industry, and energy/transport) (0.7 Tg N yÿ1 ) sources. Natural terrestrial N2 O emissions are from natural soils (6.6 Tg N yÿ1 ). Anthropogenic terrestrial emissions total 3.2 Tg N yÿ1 and include a wide range of sources: arable lands (1.0 Tg yÿ1 ), N excretion from animals (1.0 Tg N yÿ1 ), biomass burning (0.5 Tg N yÿ1 ), agricultural waste burning (0.1 Tg N yÿ1 ), industry (0.5 Tg N yÿ1 ) and combustion processes (0.2 Tg N yÿ1 ) (Bouwman et al., 1995; obtained through EDGAR (Olivier et al., 1996)). Since Bouwman et al. (1995) compiled the gridded inventory, some additional sources have been quanti®ed on a global scale but as yet are not available in gridded inventories. These sources include animal
Fig. 7. Comparison of N2 O emissions from aquatic (rivers, estuaries, continental shelves and oceans) and terrestrial systems. The anthropogenic and natural component of each is also indicated.
waste management systems, agricultural soil emissions due to crop residues and biological N2 ®xation, and sewage treatment plants. These sources may amount to over 2 Tg N yÿ1 at present (Mosier et al., 1998). The industrial and energy sources included in EDGAR are about 1 Tg N yÿ1 lower than more recent IPCC estimates by Prather et al. (1995). Based on the above, approximately 67% (6.6 Tg N yÿ1 ) of total terrestrial N2 O emissions (9.8 Tg N yÿ1 ) are considered natural and 33% (3.2 Tg N yÿ1 ) anthropogenic (Fig. 7). This is similar to the distribution of total aquatic N2 O emission, with approximately 75% considered natural and 25% anthropogenic. Globally, then, about one third of N2 O emissions are anthropogenic (more than 4 Tg N yÿ1 ). This is consistent with the observed increase in atmospheric N2 O which indicates that net additions to the atmosphere amounted to 4±5Tg N yÿ1 in the period 1977±1987 (Khalil and Rasmussen, 1992), although we know that we are missing several sources as discussed above. It should be noted, however, that including the sources for which no gridded inventories are yet available, would increase the relative contribution of anthropogenic sources. It also should be realized that the net additions of nitrous oxide to the atmosphere may be smaller than the total anthropogenic source, since emissions from natural systems may have decreased as a result of human activities (Kroeze et al., 1999). Geographical distribution of natural and anthropogenic biogenic N2 O emissions: The above comparisons were at the global scale. The geographic distribution of aquatic and terrestrial N2 O emissions,including natural and anthropogenic, was obtained by summing the gridded (1° ´ 1°) emissions inventories by latitudinal zones (0°±20°, 20°±45°, 45°±66°, 66°±90°) in each hemisphere. The latitudinal distribution of aquatic N2 O emissions diers from the biogenic terrestrial emissions (Fig. 8A). Overall, the aquatic sources are somewhat more evenly distributed among these latitudinal zones compared to the terrestrial sources, which are concentrated between 20°S and 45°N. Additional patterns emerge from an examination of natural and anthropogenic emissions from aquatic and terrestrial systems. Natural terrestrial emissions dominate in the tropical latitudes (20°S to 20°N) while over half of the anthropogenic terrestrial sources are in the mid-latitudes of the Northern Hemisphere (20° to 66°N) (Fig. 8B and C). Natural aquatic sources are relatively evenly distributed among the latitudinal zones, however anthropogenic aquatic emissions show a de®nite peak in the mid-latitudes of the Northern Hemispheric (20° to 66°N) as do the anthropogenic terrestrial emissions. The similarity in the latitudinal distribution of anthropogenic terrestrial and aquatic sources is reasonable. Anthropogenic N inputs to the terrestrial environment not only drive anthropogenic N2 O emissions from land, but also are primarily
S.P. Seitzinger et al. / Chemosphere ± Global Change Science 2 (2000) 267±279
277
Fig. 8. Comparison of the latitudinal distribution of biogenic N2 O emissions from terrestrial and aquatic systems: (A) totals; (B) natural; (C) anthropogenic emissions.
responsible for anthropogenic N2 O emissions from aquatic ecosystems. This is because a substantial portion of the anthropogenic N applied to watersheds enters rivers, their estuaries, and continental shelves, resulting in the production of N2 O (Fig. 1). Our analysis indicates that most emissions of N2 O take place in the Northern Hemisphere (63%), with largest contributions from the mid-latitudes; 37% of the global emissions are from the Southern Hemisphere. This is consistent with Prinn et al. (1990) who concluded from analysis of atmospheric trends between 1978 and 1988 that emissions into the 90°±30°N, 30°±0°N, 0°±30°S and 30°±90°S latitudinal belts contribute 22±34%, 32±39%, 20±29%, and 11±15%, respectively, to the global emis-
sions. Bouwman et al. (1995) also estimated, for their case 4b, that 60% of the total emissions are from the Northern Hemisphere and 40% from the Southern. Adding the emissions from rivers, estuaries and continental shelves to the Bouwman et al. inventory slightly improved the comparison to the Prinn et al., mid-point estimate for the North (63%)±South (37%) distribution as reported in Bouwman et al. (1995). Our estimated total global source (15 Tg N yÿ1 ) is higher than that of Prinn et al. (1990) (13 Tg N yÿ1 ), which is also consistent with the fact that Prinn et al. (1990) used a larger atmospheric lifetime for N2 O (about 170 year) than was used in more recent studies (120 years; Prather et al., 1995). As a result, Prinn et al. (1990) may have underestimated total global emissions.
278
S.P. Seitzinger et al. / Chemosphere ± Global Change Science 2 (2000) 267±279
N2 O emissions linked to N inputs from terrestrial sources versus N inputs from aquatic sources: A ®nal perspective from which to examine biogenic N2 O emissions is to consider whether the N driving the N2 O emissions in an ecosystem was originally from land or from aquatic sources. Aquatic N sources include biological N2 -®xation in aquatic ecosystems. In the framework of the current analysis, N2 O emissions associated with aquatic N sources include all ocean N2 O emissions (3.5 Tg N yÿ1 ) and that portion (55%) of continental shelf emissions that are associated with onwelling or advection of ocean N (0.3 Tg N yÿ1 ; Fig. 6) (Table 3). N2 O emissions attributed to a terrestrial N source include all biogenic N2 O from terrestrial ecosystems (2.5 Tg N yÿ1 anthropogenic, 6.6 Tg N yÿ1 natural soils), rivers (1.1 Tg N yÿ1 ) and estuaries (0.22 Tg N yÿ1 ) plus that portion of continental shelf emissions associated with atmospheric N deposition and river export (0.3 Tg N yÿ1 ) (Tables 1±3; Fig. 6). Thus, of the 14.5 Tg of biogenic N2 O±N emitted per year, approximately 75% (10.7 Tg N yÿ1 ) is due to terrestrial N and 25% (3.8 Tg N yÿ1 ) is from oceanic N sources.
4. Concluding remarks
Acknowledgements We would like to thank L. Bouwman and J. Olivier for assistance with obtaining the terrestrial and ocean data bases from EDGAR. This work was supported in part with funding from NOAA NJ Sea Grant (SPS) (NJSG-99-423), NOAA Coastal Ocean Program (SPS) and Wageningen University (CK). References
It should be kept in mind that the above analyses are based on N2 O emissions inventories currently available in gridded format. As noted above, there are a number of missing sources (Mosier et al., 1998), as well as more recent estimates of abiotic sources (Prather et al., 1995). Emissions from all sources have considerable uncertainty associated with them. As we improve our understanding of factors controlling N2 O emissions in ecosystems worldwide, most certainly our estimates will need to be reassessed. As anthropogenic activities associated with the use and cycling of N in the environment increase in the fu-
Table 3 N2 O emissions (Tg N yÿ1 ) linked to N inputs from terrestrial sources versus N inputs from aquatic sourcesa
a
ture, it will be particularly important to understand factors controlling N2 O emissions so that eective strategies can be developed to control emissions. The relatively large magnitude of freshwater and coastal marine N2 O emissions in combination with their large anthropogenic component indicates the need to obtain better quanti®cation of these sources and of the factors controlling them. This is illustrated by recent estimates of future N2 O emissions from rivers, estuaries and continental shelves which indicate that emissions from these sources alone may increase from 2 Tg N yÿ1 in 1990 to 5 Tg N yÿ1 by 2050 due to projected increases in N inputs to these aquatic ecosystems from anthropogenic activities (Kroeze and Seitzinger, 1998).
Region
N2 O from aquatic N
Oceans Continental shelves Estuaries Rivers Terrestrial soils Natural and anthropogenic Total
3.5 0.3
N2 O from terrestrial N 0.3 0.2 1.1 9.1
3.8
10.7
Inventories from Nevison et al. (1995), Bouwman et al. (1995) and Seitzinger and Kroeze (1998) (see text for details).
Bange, H.W., Rapsomanikis, S., Andreae, M.O., 1996. Nitrous oxide in coastal waters. Global Biogeochem. Cycles 10 (1), 197±207. Bouwman, A.F., Fung, I., Matthews, E., John, J., 1993. Global analysis of the potential for N2 O production in natural soils. Global Biogeochem. Cycles 7, 557±597. Bouwman, A.F., Hoek, K.W., Olivier, J.G.J., 1995. Uncertainties in the global sources distribution of nitrous oxide. J. Geophys. Res. 100, 2785±2800. Christensen, J.P., Murray, J.W., Devol, A.H., Codispoti, L.A., 1987. Denitri®cation in continental shelf sediments has major impact on the oceanic nitrogen budget. Global Biogeochem. Cycles 1, 97±116. Dentener, F.J., Crutzen, P.J., 1994. A three-dimensional model of the global ammonia cycle. J. Atmos. Chem. 19, 331±369. Edwards, M.H., 1986. Digital Image Processing of Local and Global Bathymetric Data. Department of Earth and Planetary Sciences, Washington University, USA. Erickson, D.J., 1993. Stability dependent theory for air±sea gas exchange. J. Geophys. Res. 98, 8271±8288. Food and Agriculture Organization Yearbook. Fertilizer, 1990, FAO Statistics Series No. 100, Vol. 40. Galloway, J.N., Schlesinger, W.H., Levy II, V., Michaels, A., Schnoor, J.L., 1995. Nitrogen ®xation: anthropogenic enhancement ± environmental response. Global Biogeochem. Cycles 9 (2), 235±252. Howarth, R.W., Billen, G., Swaney, D., Townsend, A., Jaworski, N., Downing, J.A., Elmgren, R., Caraco, N., Lajtha, K., 1996. Regional nitrogen budgets and riverine N & P ¯uxes for the drainages to the North Atlantic Ocean: natural and human in¯uences. Biogeochemistry 35, 75±139.
S.P. Seitzinger et al. / Chemosphere ± Global Change Science 2 (2000) 267±279 Jensen, H.B., Jorgensen, K.S., Sorensen, J., 1984. Diurnal variation of nitrogen cycling in coastal, marine sediments. II. Nitrous oxide emission. Marine Biology 83, 177±183. Khalil, M.A.K., Rasmussen, R.A., 1992. The global sources of nitrous oxide. J. Geophys. Res. 97 (D13), 14651±14660. Kroeze, C., Seitzinger, S.P., 1998. Nitrogen inputs to rivers, estuaries and continental shelves and related nitrous oxide emissions in 1990 and 2050: a global model. Nutrient Cycling Agroecosyst. 52, 195±212. Kroeze, C., Mosier, A., Bouwman, L., 1999. Closing the global N2 O budget: a retrospective analysis 1500±1994. Global Biogeochem. Cycles 13 (1), 1±8. Lerner, J., Matthews, E., Fung, I., 1998. Methane emission from animals: a global high resolution database. Global Biogeochem. Cycles 3, 281±285. Matthews, E., 1994. Nitrogenous fertilizers: global distribution of consumption and associated emissions of nitrous oxide and ammonia. Global Biogeochem. Cycles 8 (4), 411±439. Meybeck, M., 1982. Carbon, nitrogen, and phosphorus transport by world rivers. Amer. J. Sci. 282, 401±450. Mosier, A., Kroeze, C., Nevison, C., Oenema, O., Seitzinger, S., Cleemput, O., 1998. Closing the global atmospheric N2 O budget: nitrous oxide emissions through the agricultural nitrogen cycle. Nutrient Cycling Agroecosyst. 52, 225±248. Nevison, C.D., 1994. A model analysis of the spatial distribution and temporal trends of nitrous oxide sources and sinks. Ph.D. Dissertation, Stanford University, 218 pp. Nevison, C.D., Weiss, R.F., Erickson, D.J., 1995. Global oceanic emissions of nitrous oxide. J. Geophys. Res. 100, 15809±15820. Nishio, T., Koike, I., Hatorri, A., 1983. Estimates of denitri®cation and nitri®cation in coastal and estuarine sediments. Appl. Environ. Microbiol. 45, 444±450. Nixon, S.W., Ammerman, J., Atkinson, L., Berounsky, V., Billen, G., Boicourt, W., Boynton, W., Church, T., DiToro, D., Elmgren, R., Garber, J., Giblin, A., Owens, R.J., Pilson, M.E.Q., Seitzinger, S., 1996. The fate of nitrogen and phosphorus at the land-sea margin of the North Atlantic Ocean. Biogeochemistry 35, 141±180. Olivier, J.G.J., Bouwman, A.F., van der Maas, C.W.M., Berdowski, J.J.M., Veldt, C., Bloos, J.P.J., Visschedijk, A.J.H., Zandveld, P.Y.J., Haverlag, J.L., 1996. Description of EDGAR Version 2.0: A set of global emission inventories of greenhouse gases and ozone-depleting substances for all anthropogenic and most natural sources on aa per
279
country basis and on 1x1 degree grid. RIVM Report no. 771060002/TNO-MEP Report no R96/119, National Institute for Public Health and the Environment, Bilthoven, The Netherlands. Pilson, M.E., Seitzinger, S.P., 1996. Areas of shallow water in the North Atlantic. Biogeochemistry 35, 260±264. Pilson, M.E.Q., 1998. An Introduction to the Chemistry of the Sea. Prentice-Hall, New York. Prather, M., Derwent, R., Ehhalt, D., Fraser, P., Sanhueza, E., Zhou, X., 1995. Other trace gases and atmospheric chemistry. In: Houghton, J.T., Meira Filho, L.G., Bruce, J., Lee, H., Callander, B.A., Haites, E., Harris, N., Maskell, K. (Eds.). Climate Change 1994: Radiative Forcing of Climate Change and an Evaluation of the IPCC IS92 Emission Scenarios, the Intergovernmental Panel on Clmiate Change, Cambridge University Press, pp. 77±126. Prinn, R., Cunnold, D., Rasmussen, R., Simmonds, P., Aleya, F., Crawford, A., Fraser, P., Rosen, R., 1990. Atmospheric emissions and trends of nitrous oxide deduced from 10 years of ALE-GAGE data. J. Geophys. Res. 95, 18369± 18385. Seitzinger, S.P., 1988. Denitri®cation in freshwater and coastal marine ecosystems: ecological and geochemical signi®cance. Limnol. Oceanogr. 33 (4), 702±724. Seitzinger, S.P., Giblin, A.E., 1996. Estimating denitri®cation in North Atlantic Continental Shelf sediments. Biogeochemistry 35, 235±260. Seitzinger, S.P., Kroeze, C., 1998. Global distribution of nitrous oxide production and N inputs in freshwater and coastal marine ecosystems. Global Biogeochem. Cycles 12, 93±113. Van Aardenne, J.A., Kroeze, C., Hordijk, L., 1998. Analysis of the uncertainties in the IPCC default method for estimating N2O emissions from agricultural soils. In: Chan, K., Tarantola, S., Campolongo, F. (Eds.). Proceedings of SAMO, Second International Symposium on Sensitivity Analysis of Model Output, ISIS-SMA, EC Joint Research Centre, Ispra, Italy. Sybil SeitzingerÕs current research includes studies of: denitri®cation in rivers, estuaries and continental shelves; dissolved organic nitrogen inputs and bioavailability in aquatic ecosystems; and global modeling of N transport by world rivers and associated nitrous oxide production. Anthropogenic impacts are key components of her studies. She also enjoys windsur®ng, ice climbing, snow boarding, white water kayaking, and traveling.
Global distribution of N2O emissions from aquatic systems: natural emissions and anthropogenic eects Sybil P. Seitzinger a,*, Carolien Kroeze b, Renee V. Styles a a
Institute of Marine and Coastal Sciences, Rutgers University, 71 Dudley Road, New Brunswick, NJ 08901, USA b Institute of Environment and Climate Research, Wageningen University, Wageningen, Netherlands Received 9 July 1999; accepted 12 October 1999
Abstract Context Abstract: Atmospheric concentrations of nitrous oxide, a greenhouse gas, are increasing due to human activities. Our analysis suggests that a third of global anthropogenic N2 O emission is from aquatic sources (rivers, estuaries, continental shelves) and the terrestrial sources comprise the remainder. Over 80% of aquatic anthropogenic N2 O emissions are from the Northern Hemisphere mid-latitudes consistent with the geographic distribution of N fertilizer use, human population and atmospheric N deposition. These N inputs to land have increased aquatic as well as terrestrial anthropogenic N2 O emissions because a substantial portion enters aquatic systems and results in increased N2 O production. Thus, wise management of N in the terrestrial environment could help reduce/control both aquatic and terrestrial N2 O emissions. Main Abstract: The global distribution of N2 O emissions from rivers, estuaries, continental shelves, and oceans are compared to each other, and to terrestrial emissions, using existing gridded inventories. Rivers, estuaries and continental shelves (1.9 Tg N yÿ1 ) account for about 35% of total aquatic N2 O emissions; oceanic emissions comprise the remainder. Oceanic N2 O emissions are approximately equally distributed between the Northern and Southern Hemispheres; however, over 90% of emissions from estuaries and rivers are in the Northern Hemisphere. N2 O emissions from rivers, estuaries, and continental shelves combined equal oceanic emissions in both the 20°±45°N and 45°±66°N latitudinal zones. Over 90% of river and estuary emissions are considered anthropogenic (1.2 Tg N yÿ1 ); only 25% of continental shelf emissions are considered anthropogenic (0.1 Tg N yÿ1 ); oceanic emissions are considered natural. Overall, approximately one third of both aquatic and of terrestrial emissions are anthropogenic. Natural terrestrial emissions are highest in tropical latitudes while natural aquatic emissions are relatively evenly distributed among latitudinal zones. Over half of both the anthropogenic terrestrial and aquatic emissions occur between 20° and 66°N. Anthropogenic N inputs to the terrestrial environment drive anthropogenic N2 O emissions from both land and aquatic ecosystems, because a substantial portion of the anthropogenic N applied to watersheds enters rivers, estuaries and continental shelves. Ó 2000 Elsevier Science Ltd. All rights reserved. Keywords: Nitrous oxide; Aquatic; Rivers; Estuaries; Continental shelves; Anthropogenic; Global emissions; Geographic distribution
1. Introduction *
Corresponding author. Tel.: +1-732-932-6555; fax: +1-732932-8578. E-mail addresses: [email protected] (S.P. Seitzinger), [email protected] (C. Kroeze), [email protected] (R.V. Styles).
Total global N2 O emissions include a wide range of sources from terrestrial and aquatic systems, with both natural and anthropogenic components (e.g., Khalil and Rasmussen, 1992; Matthews, 1994; Bouwman et al., 1995; Bange et al., 1996). A clear understanding of the
1465-9972/00/$ - see front matter Ó 2000 Elsevier Science Ltd. All rights reserved. PII: S 1 4 6 5 - 9 9 7 2 ( 0 0 ) 0 0 0 1 5 - 5
268
S.P. Seitzinger et al. / Chemosphere ± Global Change Science 2 (2000) 267±279
magnitude of the various natural and anthropogenic sources and their geographic distribution is necessary to quantify factors contributing to the current increases in atmospheric N2 O concentrations, to predict future potential increases, and to develop realistic and costeective strategies to reduce future emissions. There have been a number of analyses of the magnitude and global distribution of terrestrial and deep ocean sources (e.g., Nevison, 1994; Bouwman et al., 1995); however only recently have such analyses been done for freshwater and coastal marine systems (Seitzinger and Kroeze, 1998). N2 O emissions from freshwater and coastal marine ecosystems are estimated to amount to 1.9 Tg N yÿ1 (Seitzinger and Kroeze, 1998), which is similar in magnitude to estimated N2 O emissions from a number of other sources, including arable land, grasslands, and animal excreta (Bouwman et al., 1995). Much of the N2 O emission from these aquatic systems is anthropogenic and thus contributes to the current increase in atmospheric N2 O. In addition, large increases in future N2 O emissions from these aquatic systems are predicted as a result of increases in human activities (Kroeze and Seitzinger, 1998). Bouwman et al. (1995) presented an analysis of the total and latitudinal distribution of most of the known terrestrial sources and Nevison et al. (1995) has done so with oceanic sources. We now extend that analysis by adding freshwater and coastal marine N2 O emissions. In the current paper we compare: (1) N2 O emissions from rivers, estuaries and continental shelves with N2 O emissions from oceanic regions, (2) anthropogenic and natural emissions from aquatic systems, and (3) aquatic and terrestrial N2 O emissions, both natural and anthropogenic. These comparisons are done at a global scale as well as on a latitudinal basis to provide insight into the geographic distribution of the dierent sources. Gridded (1° ´ 1°) estimates of N2 O emissions from published studies are used in these analyses. 2. Methods We calculated N2 O emissions for the following latitudinal zones for the northern and southern hemisphere: 0°±20°, 20°±45°, 45°±66° and 66°±90°. Emissions from oceans: We obtained the gridded (1° ´ 1°) global N2 O emission inventory for ocean emissions from Nevison (1994) through the Emission Database for Global Atmospheric Research (EDGAR) (Olivier et al., 1996). The EDGAR database had been converted from the original 2.8° ´ 2.8° grid of Nevison (1994) to a 1° ´ 1° grid, which resulted in a loss of 0.2 Tg N yÿ1 from the original 3.8 Tg N yÿ1 estimate (Bouwman et al., 1995). That database (1° ´ 1°) is based on NevisonÕs ``best'' estimate emissions which she cal-
culated using the water±air ¯ux calculation of Erickson (1993). It is also available from the Global Emissions Inventory Activity (GEIA) web site (http://blueskies.sprl.umich.edu/geia). We excluded from Nevison's database the N2 O emissions from the area that is considered continental shelf by Seitzinger and Kroeze (1998), i.e., the gridcells with an average depth <200 m. The totals for N2 O emissions from oceans by latitudinal zones then were calculated by summing the N2 O emissions from all gridcells with an average depth >200 m. Emissions from continental shelves: We used the gridded (1° ´ 1°) global N2 O inventory from Seitzinger and Kroeze (1998). The totals for latitudinal zones were calculated by summing the emissions from the continental shelf gridcells within these zones. Emissions from estuaries: We used the gridded (1° ´ 1°) global N2 O inventory from Seitzinger and Kroeze (1998). The totals for latitudinal zones were calculated by summing the emissions from the estuarine gridcells within these zones. Emissions from rivers: We used the gridded (1° ´ 1°) global N2 O inventory from Seitzinger and Kroeze (1998). The total emissions for latitudinal zones were calculated by summing the emissions from the watershed gridcells within a latitudinal zone. This diers from the approach used by Seitzinger and Kroeze (1998) to assign river N2 O emissions to latitudinal zones; they assigned the N2 O emissions from an entire watershed to the latitude of the river mouth for that watershed. Terrestrial emissions: We used the gridded (1° ´ 1°) global N2 O emission inventory for terrestrial emissions from dierent source categories from Bouwman et al. (1993, 1995) which we obtained through the EDGAR (Olivier et al., 1996). It is also available through the GEIA website (http://blueskies.sprl.umich.edu/geia). The terrestrial emission source categories distinguished by Bouwman et al. (1995) include natural soils, arable land, animal excreta, biomass burning, agricultural waste burning, post-clearing eects of deforestation, industry, energy and fuelwood combustion (Bouwman et al., 1995). In the EDGAR database several of these sources have been broken down into subcategories, totalling 16 sources of N2 O. Emissions from a 17th subcategory, fossil fuel combustion from air transportation, were not available as a gridded database from EDGAR, so they were not included in our terrestrial totals. N2 O emissions from this source are 0.003 Tg N yÿ1 , which is equivalent to 0.09% of the total anthropogenic terrestrial emissions (Olivier et al., 1996). We used the aquatic inventories identi®ed above in the current analysis because as far as we are aware they are the only gridded inventories available for rivers, estuaries, continental shelves or oceans. Details of the methods used to estimate emissions from each of these sources and uncertainties in each inventory are discussed in the original publications (Nevison et al., 1995; Kroeze
S.P. Seitzinger et al. / Chemosphere ± Global Change Science 2 (2000) 267±279
and Seitzinger, 1998; Seitzinger and Kroeze, 1998). There have been a number of estimates of terrestrial N2 O emission from particular sources (e.g., Matthews, 1994; Nevison, 1994) and some additional sources or modi®cations to the above estimates (discussed below) since Bouwman et al. (1995). We used the inventories of Bouwman et al. (1995) because of their relative completeness and availability in gridded format. The reader is referred to Bouwman et al. (1993, 1995) for discussion of the uncertainties in those terrestrial inventories.
3. Results and discussion 3.1. Aquatic N2 O emissions N2 O emissions from rivers, estuaries and continental shelves: Global watersheds receive inputs of nitrogen (N) from a variety of anthropogenic sources (Fig. 1). The three anthropogenic sources of new ®xed N are synthetic fertilizer, combustion of fossil fuels and biological N2 ®xation associated with cultivated crops. Together, these sources contribute an amount of N (about 90±130 Tg N yÿ1 ) similar to the natural biological ®xation of N2 in non-agricultural soils (140 Tg N yÿ1 ) (Galloway et al., 1995). A considerable amount of the anthropogenic N inputs to terrestrial ecosystems eventually enters rivers through leaching, runo and direct discharge. A portion of this N in rivers is transported to coastal marine eco-
Fig. 1. Schematic of N inputs to aquatic ecosystems. N inputs to rivers and estuaries are linked to N inputs to the surrounding watersheds from both natural and anthropogenic sources. Continental shelves receive N inputs from both terrestrial sources (river/estuarine export and atmospheric deposition) and from oceanic (e.g., onwelling, advection) sources. Subsequent cycling of N through nitri®cation and denitri®cation pathways results in N2 O emissions from these aquatic systems.
269
systems (estuaries and continental shelves). The biological cycling of natural and anthropogenic nitrogen through these aquatic ecosystems results in emissions of N2 O during both nitri®cation and denitri®cation. The global geographic distribution of dissolved inorganic N (DIN) export by world rivers to coastal marine ecosystems and associated N2 O emissions in rivers and coastal marine ecosystems were estimated in Seitzinger and Kroeze (1998). The detailed methodology is presented in Seitzinger and Kroeze (1998). A brief description of the approach is presented here. DIN input to and export by rivers was modeled as a function of N inputs to the watersheds of rivers from synthetic fertilizer use, atmospheric NOy deposition, human sewage, and water runo. (Input and export refer to ¯uxes.) (N input from the watershed areas below the mouth of a river, for example from cities located on estuaries below river mouths, also was included in the estimated river export.) N2 O emissions in rivers and their estuaries were estimated as a function of denitri®cation and nitri®cation using the emission factors described below; denitri®cation and nitri®cation were estimated as a function of external N inputs to the rivers or estuaries. Emission factors of N2 O for nitri®cation and denitri®cation rates were 0.3% for rivers and estuarine gridcells, except in gridcells with external N inputs exceeding 10 kg N haÿ1 yÿ1 ; for these, an emission factor of 3% was used. These emissions factors are based on a variety of studies which suggest that the ratio of N2 O:N2 ¯uxes is generally within the range of 0.1±0.5%, with ratios up to 6% in heavily polluted sediments (e.g., Nishio et al., 1983; Jensen et al., 1984; Seitzinger, 1988). N2 O emissions in continental shelf regions also were estimated as a function of nitri®cation and denitri®cation, however with a lesser amount of detail than for rivers and estuaries (Seitzinger and Kroeze, 1998). N2 O production associated with denitri®cation was estimated by extrapolating previously estimated denitri®cation rates for three latitude zones (0°±20°, 20°±45° and 45°± 90°) in the North Atlantic (Seitzinger and Giblin, 1996) to shelf areas in those latitudes in other ocean basins. N2 O production associated with pelagic nitri®cation was estimated by applying the depth average rate of nitri®cation measured in shelf waters of the western US to the global shelf area. Continental shelves were de®ned as areas with water depths less than 200 m using the USDOC/NOAA ETOPO-5 bathymetry database (Edwards, 1986; Pilson and Seitzinger, 1996). An emission factor of 0.3% was applied for N2 O for nitri®cation and denitri®cation rates in continental shelf regions. Relative magnitude of all aquatic sources: Model predicted N2 O emissions from global rivers (1.1 Tg N yÿ1 ), estuaries (0.2 Tg N yÿ1 ) and continental shelves (0.6 Tg N yÿ1 ) total 1.9 Tg N2 O-N yÿ1 (Seitzinger and Kroeze, 1998) (Fig. 2). This substantially increases the only other gridded inventory of global N2 O emissions
270
S.P. Seitzinger et al. / Chemosphere ± Global Change Science 2 (2000) 267±279 Rivers 1.1
Estuaries 0.2 Continental Shelves 0.6
Ocean 3.5
Fig. 2. Comparison of global emissions of N2 O from all aquatic ecosystems (no global emissions estimates for lakes). Units: Tg N yÿ1 (emissions estimates from Seitzinger and Kroeze, 1998 and Nevison et al., 1995).
from aquatic ecosystems which is that from oceanic regions (3.5 Tg N yÿ1 ; NevisonÕs (1994) inventory minus emissions from continental shelf gridcells). Global emissions from lakes have not been estimated. Overall the N2 O emissions from freshwater and coastal systems (1.9 Tg N yÿ1 ) account for about 35% of the total aquatic emissions (5.4 Tg N yÿ1 ) (Fig. 2). The uncertainties in the freshwater and coastal marine N2 O emissions estimates are considerable. For rivers, estuaries and continental shelves (1.9 Tg N yÿ1 ), the estimated uncertainty range is 0.9±9.0 Tg N yÿ1 , based on uncertainties in emission factors only (Seitzinger and Kroeze, 1998). The emission estimates were derived by using dierent emission factors, which depended on the external N inputs (0.3 or 3% of nitri®cation and denitri®cation rates; see above). A sensitivity analysis indicated that a twofold increase or decrease in the threshold N input value used to determine the emission factor would not markedly alter the overall conclusions of the study by Seitzinger and Kroeze (1998), given the other model uncertainties. Their analysis furthermore indicated that the calculated emissions from rivers are less sensitive than emissions from estuaries to changes in the threshold N input. Unfortunately, there are few rivers or estuaries where N2 O emissions have been measured in conjunction with N input over the temporal (yearly) and spatial (entire river or estuary) scales necessary to compare with the model output. A comparison of the modeled N2 O emissions from the Amazon River and Tamar and Narragansett estuaries indicated that the modeled emissions agree within a factor of 2±5. However, additional comparisons across a range of rivers and estuaries are needed to validate the modeled N2 O ¯uxes. The estimates for aquatic emissions of nitrous oxide are likely to be more uncertain than emissions from terrestrial systems, primarily because there are many more studies of N2 O emissions in terrestrial systems. Van Aardenne et al. (1998) analyzed the
uncertainties in agricultural emissions of nitrous oxide from terrestrial systems for the Netherlands, following the methodology in Mosier et al. (1998). Van Aardenne et al. (1998) also estimated aquatic N2 O emissions associated with external N inputs (leaching) to aquatic systems from agricultural sources using the methodology in Mosier et al. (1998) which is similar to that of Seitzinger and Kroeze (1998). Van Aardenne et al. (1998) concluded from a Monte Carlo analysis, that of the dierent model parameters analyzed, relatively large contributions to the total uncertainty were due to uncertainties in the emission factors from aquatic systems and agricultural ®elds, and in the estimated fraction of agricultural nitrogen subject to leaching and runo. The latitudinal distribution of aquatic sources: There are obvious patterns in the latitudinal distribution of N2 O emissions from aquatic sources (Fig. 3; Table 1). Oceanic N2 O emissions are approximately equally distributed between the northern (42%) and southern (58%) hemispheres. Approximately 70% of the total continental shelf emissions are in the northern hemisphere. However, N2 O emissions from estuaries and rivers are almost exclusively in the northern hemisphere (96% and 98%, respectively). A more detailed look at the distribution of emissions (Tg N yÿ1 ) by latitudinal zones (0°±20°, 20°±45°, 45°±66° and 66°±90° in Northern and Southern Hemispheres) indicates further dierences in the pattern among aquatic sources. For example, the total amount of N2 O emitted from the oceans in each of the ®ve latitudinal zones between 45°S and 66°N is similar, with total oceanic emissions between 45°S and 66°S being at least twice as great (Fig. 3). Continental shelf emissions also are similar in each of the ®ve latitudinal zones between 45°S and 66°N, although there is a slight trend of decreasing emissions going from north to south. In contrast, N2 O emissions from rivers or estuaries show considerable variation among latitudinal zones with the 20°±45°N and 45°±66°N zones having the highest total emissions. N2 O emissions from rivers, estuaries, and continental shelves combined equal oceanic emissions in both the 20°±45°N and 45°±66°N latitudinal zones. These dierences in latitudinal distributions among aquatic sources are related to a variety of factors. Oceanic N2 O ¯uxes were calculated by Nevison et al. (1995) by coupling sea±air gas transfer coecient ®elds to a map of surface water N2 O supersaturations obtained from cruise data. In general, the latitudinal distribution of surface N2 O re¯ects both wind velocities and the amount of mixing between N2 O rich deep water and surface waters (Nevison et al., 1995). An overall view of the latitudinal distribution of ocean N2 O emission per unit area (¯uxes) was obtained by combining the ocean area (Fig. 4(A)) with the ocean N2 O emission for each latitudinal zone (Fig. 3). A pattern of decreasing N2 O
S.P. Seitzinger et al. / Chemosphere ± Global Change Science 2 (2000) 267±279
271
Fig. 3. Comparison of the latitudinal distribution of total N2 O emissions from rivers, estuaries, continental shelves and oceans Units: Tg N yÿ1 .
emission per unit area of ocean is apparent between 66°N and 45°S (Fig. 4(C)). In the Southern Hemisphere between 45°±66°S there is a large increase in rate per unit area (¯ux) due to upwelling of N2 O rich deep water (Nevison, 1994; Nevison et al., 1995). There is a strong pattern of decreasing continental shelf area going from the Northern to the Southern Hemisphere (Fig. 4(B)); however, the N2 O ¯ux, as de®ned in the current calculations, does not vary greatly among latitudinal zones. This results in the general decrease in total N2 O emitted from continental shelf regions per latitudinal zone going from north to south (Fig. 3). The latitudinal distribution of river and estuary N2 O emissions (Fig. 3) is consistent with the distribution of N inputs to watersheds from anthropogenic activities such as fertilizer use (FAO, 1990 modi®ed by Bouwman et al., 1995), human population density (Lerner et al., 1988 updated to 1990 by Bouwman) and atmospheric NOy deposition (Dentener and Crutzen, 1994) which are dominant in the northern hemisphere between 20°N and 66°N (Fig. 5(A)±(C)). Relative magnitude of natural and anthropogenic aquatic N2 O emissions: In the current analysis we attribute N2 O emissions to natural or anthropogenic sources depending on the origin (natural or anthropogenic) of the N source to the aquatic system that is nitri®ed or denitri®ed. N inputs to oceanic regions are considered to be essentially all natural in that they are primarily from N2 -®xation in the oceans; thus N2 O emissions in the oceans linked to nitri®cation and denitri®cation of that N are considered natural (Table 2). Two potential additional sources of N to oceanic regions are atmospheric deposition and rivers. All N inputs from rivers are likely removed by denitri®cation and burial in
the sediments of estuaries and continental shelves (Christensen et al., 1987; Seitzinger and Giblin, 1996) and thus do not contribute to N inputs to oceanic regions. Anthropogenic NOy deposition to coastal and oceanic regions is estimated at 9 Tg N yÿ1 (Galloway et al., 1995). However, given that anthropogenic NOy deposition is relatively small compared to N inputs from oceanic N2 -®xation (wide range of estimates but generally >80 Tg N yÿ1 ; Pilson, 1998), and that only a portion of the deposition is to oceanic regions, in this analysis it is considered negligible. N inputs to rivers are from a combination of both natural and anthropogenic sources. Knowing what component of the N in rivers is from natural or anthropogenic sources is dicult to estimate because of a paucity of measurements of N in rivers with no anthropogenic activity in their watersheds. In a recent model of DIN export by world rivers, roughly 25% (5 Tg N yÿ1 ) of a total of 20.8 Tg yÿ1 was estimated to be natural N (Seitzinger and Kroeze, 1998). This estimate was made by simply assuming that natural export of DIN is 40 kg N (kmÿ2 watershed) yÿ1 , as estimated by Meybeck (1982) based on available data for relatively unpolluted rivers globally. The actual value is likely to vary geographically (Howarth et al., 1996), however, at present a better estimate is not available. Globally, N2 O emissions from rivers are estimated to be 1.1 Tg N yÿ1 (Seitzinger and Kroeze, 1998). The emission factor used by Seitzinger and Kroeze (1998) to relate N2 O emissions to denitri®cation in rivers varies as a function of N inputs to rivers (see above). Therefore, to estimate natural N2 O emissions from rivers, we assumed a natural watershed DIN yield of 40 kg N (kmÿ2 watershed) yÿ1 (Meybeck, 1982) and applied the approach of Seitzinger
a
278 193 0 17.8 431 920
0.053 0.001 0 0.003 28.1 28.2 0.624 28.7 149
Total anthropogenic terrestriala
Total terrestrial
Total terrestrial and aquatic
4151
2599
1409
560 435 35.5 46.9 1190 2267
31.0 107 194 332
579 157 145 671 1552
45°N±20°N
Total anthropogenic terrestrial includes all terrestrial sources except natural soils.
1956
1195
764
2.91 43.2 229 275
354 113 54.8 239 761
66°N±45°N
0.003 0.109 0.455 0.567
16.0 93.6 1.29 8.83 120
90°N±66°N
3330
2588
534
80.0 181 222 18.4 2054 2555
16.7 5.30 10.7 32.7
525 90.2 13.0 114 742
20°N±0°
N2 O emissions by latitudinal zones (Gg N yÿ1 )
Terrestrial Abiogenic Biofuel Energy/transport Insustrial processes Total abiogenic Biogenic Arable land Animal excretion Biomass burning Agricultural waste burning Natural soils Total biogenic
Aquatic Ocean Cont. shelf Estuaries Rivers Total aquatic
Source
3060
2613
296
25.4 88.6 162 5.34 2317 2598
7.37 1.5 5.4 14.3
342 83.3 7.15 14.3 447
0°±20°S
1299
767
199
18.6 118 32.7 5.02 568 742
2.21 5.94 16.9 25.0
452 75.1 1.39 3.57 532
20°S±45°S
1225
13.1
6.51
0.111 6.33 0 0.006 6.62 13.1
0.012 0.026 0.029 0.067
1193 19.2 0.003 0.362 1212
45°S±66°S
43.7
0
0
0 0 0 0 0 0
0 0 0 0
32.2 11.5 0 0 43.7
66°S±90°S
15213
9804
3210
963 1021 452 93.4 6594 9123
60.3 163 457 680
3493 643 222 1051 5409
Total
Table 1 N2 O emissions summed by latitudinal zones for all sources (Gg N yrÿ1 ). Emissions inventories for oceans from Nevison et al. (1995), other aquatic sources from Seitzinger and Kroeze (1998), and for terrestrial sources from Bouwman et al. (1995)
272 S.P. Seitzinger et al. / Chemosphere ± Global Change Science 2 (2000) 267±279
S.P. Seitzinger et al. / Chemosphere ± Global Change Science 2 (2000) 267±279
273
Fig. 4. Comparison of the latitudinal distribution of: (A) ocean area (106 km2 ); (B) continental shelf area (106 km2 ); (C) N2 O emissions for oceans and continental shelves per unit area of ocean or shelf (kg N kmÿ2 yÿ1 ).
and Kroeze (1998). Natural N2 O emissions from rivers are estimated to be approximately 5% (0.05 Tg N yÿ1 ) (Table 2). The remaining 95% of global N2 O emissions from rivers, 1.05 Tg N yÿ1 , are considered anthropogenic. In addition to DIN, organic N (both dissolved and particulate) is transported by, and transformed within, rivers; however, N2 O emissions associated with transformations of that organic N have not been estimated. N inputs to estuaries were estimated based on modeled N export by rivers. Natural N2 O emissions from estuaries (0.02 Tg N yÿ1 ; Table 2) were calculated using the above estimate of natural DIN export by rivers (5 Tg N yÿ1 ) and the approach of Seitzinger and Kroeze (1998). Anthropogenic N2 O emissions (0.20 Tg N yÿ1 ) were calculated as the dierence between the total es-
tuarine N2 O emissions (0.22 Tg N yÿ1 ) and natural emissions. Therefore, anthropogenic emissions account for 91% and natural emissions only 9% of the N2 O from estuaries. Continental shelf N2 O emissions are a combination of anthropogenic and natural. Again, we estimated the relative amount of natural and anthropogenic N2 O emissions based on the N sources to the shelf. Continental shelves receive total N (TN) inputs from both terrestrial sources and from the oceans (Fig. 6). The oceanic component (e.g., onwelling, across shelf transport) is considered to be natural as discussed above. Terrestrial N inputs to the shelf include river export (direct discharge of large rivers to shelves as well as export of N from estuaries) and atmospheric deposition directly to shelf waters. The relative magnitude of ter-
274
S.P. Seitzinger et al. / Chemosphere ± Global Change Science 2 (2000) 267±279
Fig. 5. The latitudinal distribution (for exoreic watershed areas only) of: (A) fertilizer use (Tg N yÿ1 ); (B) atmospheric NOy deposition (Tg N yÿ1 ); (C) human population density (billions).
restrial and ocean N inputs to shelves globally is not well known. However, an analysis was recently done for the North Atlantic (Seitzinger and Giblin, 1996). For the North Atlantic continental shelf, ocean sources (10±13.5 Tg N yÿ1 ) are estimated to contribute about 1.5 times as much N as terrestrial (rivers+atmosphere) sources (8.2±10.6 Tg N yÿ1 ) (Fig. 6). As noted above, ocean sources are considered to be natural N. The anthropogenic and natural components of river and atmospheric N inputs to the shelf are estimated as follows. Twenty-®ve percent of river DIN export to coastal systems globally are estimated to be natural (see above). We use this percentage to estimate the portion of total river N inputs (6.4±8.8 Tg N yÿ1 ) to the North Atlantic continental shelf that are natural (1.6±2.2 Tg N yÿ1 ) and
anthropogenic (4.8±6.6 Tg N yÿ1 ) (Fig. 6). Atmospheric N deposition (NOy, NHx, dissolved organic N) also contains both natural and anthropogenic components. Pre-industrial (i.e., natural) and current anthropogenic NOy deposition to the global oceans (3.3 and 9 Tg N yÿ1 , respectively) and to continents (4.5 and 22 Tg N yÿ1 , respectively) have been estimated by Galloway et al. (1995), however, the relative contribution of natural and anthropogenic NOy deposition to shelves was not estimated. Since continental shelves are located between the terrestrial and the oceanic regions, we assume that the relative amounts of natural (22%) and anthropogenic (78%) NOy deposition to the North Atlantic shelf are the average of that to continents (17% natural and 83% anthropogenic) and oceans (27% natural and 73%
S.P. Seitzinger et al. / Chemosphere ± Global Change Science 2 (2000) 267±279
275
Table 2 Estimated N2 O emissions (Tg N yÿ1 ) from aquatic ecosystems, including distribution of natural and anthropogenic emission (see text for assumptions) Source
Total N2 O emission
Natural N2 O emission
Anthropogenic N2 O emission
Continental shelves Estuaries Rivers Subtotal Oceans
0.6a 0.22a 1.1a 1.9 3.5b
0.4 (66%) 0.02 (9%) 0.05 (5%) 0.47 (25%) 3.5 (100%)
0.2 (34%) 0.20 (91%) 1.05 (95%) 1.42 (75%) 0 (0%)
5.4
4.0(74%)
1.4 (26%)
Total a b
Seitzinger and Kroeze (1998). Nevison et al. (1995) after omitting emissions from continental shelf gridcells.
Fig. 6. N inputs to North Atlantic continental shelf from estuarine export plus large river discharge, atmospheric deposition, and ocean inputs (e.g., onwelling, advection) (from Seitzinger and Giblin, 1996). The range of inputs from rivers/estuarine export is due to the range in estimates of N removal by denitri®cation within estuaries before N is exported to the shelf (Nixon et al., 1996). Ocean onwelling was calculated by dierence to balance N losses through burial and denitri®cation (from Seitzinger and Giblin, 1996).
anthropogenic). We also, somewhat simplistically, assumed the same distribution for NHx. Of the total NOy and NHx deposition to the North Atlantic continental shelf (1.8 Tg N yÿ1 ; Nixon et al., 1996), we estimate that 0.4 Tg N yÿ1 are natural and 1.4 Tg N yÿ1 are anthropogenic (Fig. 6). (No estimate of organic N deposition to continental shelf areas was available.) Considering ocean, atmospheric and river inputs of N to the North Atlantic shelf, and the distribution of natural and anthropogenic components in each, we estimate that about 66% of the N inputs to the shelf are from natural sources and 34% from anthropogenic sources. Following the approach for rivers, estuaries and oceans, we would then proportion the estimated total N2 O
emissions from this continental shelf region between natural and anthropogenic sources according to the proportion of natural and anthropogenic N inputs. For shelf areas outside the North Atlantic, no comparable analyses of the relative contributions of the sources (ocean, atmospheric, rivers) exist and they are likely to dier in other regions. The oceanic N contribution to continental shelves outside of the North Atlantic is especially uncertain. For lack of additional information, we have used the North Atlantic scenario to estimate the proportion of the total global N2 O emissions from continental shelves (0.6 Tg N yÿ1 ) that is from natural (66%; 0.4 Tg N yÿ1 ) and anthropogenic (34%; 0.2 Tg N yÿ1 ) N sources (Table 2).
276
S.P. Seitzinger et al. / Chemosphere ± Global Change Science 2 (2000) 267±279
Based on the above analyses, anthropogenic emissions dominate the freshwater and coastal marine N2 O emissions with approximately 75% attributed to anthropogenic N2 O emission and 25% to natural (Table 2). However, when ocean emissions are added the pattern reverses; now approximately 26% (1.4 Tg N yÿ1 ) of the total aquatic emissions are attributed to an anthropogenic N source and 74% (4 Tg N yÿ1 ) to a natural N source. 3.2. Comparison of aquatic and terrestrial N2 O emissions Overall comparison of natural and anthropogenic emissions: Bouwman et al. (1995) summarized the global source distribution of N2 O emissions from terrestrial sources and from the ocean. We now compare the updated aquatic N2 O emissions to the terrestrial N2 O emissions. Aquatic emissions account for about a third (5.4 Tg N yÿ1 ) and total terrestrial emissions about twothirds of the total estimated global N2 O emissions (15.2 Tg N yÿ1 ) (Fig. 7; Table 1). The total terrestrial emissions include both biogenic (natural soils and agriculture) (9.1 Tg N yÿ1 ) and abiogenic (biofuel, industry, and energy/transport) (0.7 Tg N yÿ1 ) sources. Natural terrestrial N2 O emissions are from natural soils (6.6 Tg N yÿ1 ). Anthropogenic terrestrial emissions total 3.2 Tg N yÿ1 and include a wide range of sources: arable lands (1.0 Tg yÿ1 ), N excretion from animals (1.0 Tg N yÿ1 ), biomass burning (0.5 Tg N yÿ1 ), agricultural waste burning (0.1 Tg N yÿ1 ), industry (0.5 Tg N yÿ1 ) and combustion processes (0.2 Tg N yÿ1 ) (Bouwman et al., 1995; obtained through EDGAR (Olivier et al., 1996)). Since Bouwman et al. (1995) compiled the gridded inventory, some additional sources have been quanti®ed on a global scale but as yet are not available in gridded inventories. These sources include animal
Fig. 7. Comparison of N2 O emissions from aquatic (rivers, estuaries, continental shelves and oceans) and terrestrial systems. The anthropogenic and natural component of each is also indicated.
waste management systems, agricultural soil emissions due to crop residues and biological N2 ®xation, and sewage treatment plants. These sources may amount to over 2 Tg N yÿ1 at present (Mosier et al., 1998). The industrial and energy sources included in EDGAR are about 1 Tg N yÿ1 lower than more recent IPCC estimates by Prather et al. (1995). Based on the above, approximately 67% (6.6 Tg N yÿ1 ) of total terrestrial N2 O emissions (9.8 Tg N yÿ1 ) are considered natural and 33% (3.2 Tg N yÿ1 ) anthropogenic (Fig. 7). This is similar to the distribution of total aquatic N2 O emission, with approximately 75% considered natural and 25% anthropogenic. Globally, then, about one third of N2 O emissions are anthropogenic (more than 4 Tg N yÿ1 ). This is consistent with the observed increase in atmospheric N2 O which indicates that net additions to the atmosphere amounted to 4±5Tg N yÿ1 in the period 1977±1987 (Khalil and Rasmussen, 1992), although we know that we are missing several sources as discussed above. It should be noted, however, that including the sources for which no gridded inventories are yet available, would increase the relative contribution of anthropogenic sources. It also should be realized that the net additions of nitrous oxide to the atmosphere may be smaller than the total anthropogenic source, since emissions from natural systems may have decreased as a result of human activities (Kroeze et al., 1999). Geographical distribution of natural and anthropogenic biogenic N2 O emissions: The above comparisons were at the global scale. The geographic distribution of aquatic and terrestrial N2 O emissions,including natural and anthropogenic, was obtained by summing the gridded (1° ´ 1°) emissions inventories by latitudinal zones (0°±20°, 20°±45°, 45°±66°, 66°±90°) in each hemisphere. The latitudinal distribution of aquatic N2 O emissions diers from the biogenic terrestrial emissions (Fig. 8A). Overall, the aquatic sources are somewhat more evenly distributed among these latitudinal zones compared to the terrestrial sources, which are concentrated between 20°S and 45°N. Additional patterns emerge from an examination of natural and anthropogenic emissions from aquatic and terrestrial systems. Natural terrestrial emissions dominate in the tropical latitudes (20°S to 20°N) while over half of the anthropogenic terrestrial sources are in the mid-latitudes of the Northern Hemisphere (20° to 66°N) (Fig. 8B and C). Natural aquatic sources are relatively evenly distributed among the latitudinal zones, however anthropogenic aquatic emissions show a de®nite peak in the mid-latitudes of the Northern Hemispheric (20° to 66°N) as do the anthropogenic terrestrial emissions. The similarity in the latitudinal distribution of anthropogenic terrestrial and aquatic sources is reasonable. Anthropogenic N inputs to the terrestrial environment not only drive anthropogenic N2 O emissions from land, but also are primarily
S.P. Seitzinger et al. / Chemosphere ± Global Change Science 2 (2000) 267±279
277
Fig. 8. Comparison of the latitudinal distribution of biogenic N2 O emissions from terrestrial and aquatic systems: (A) totals; (B) natural; (C) anthropogenic emissions.
responsible for anthropogenic N2 O emissions from aquatic ecosystems. This is because a substantial portion of the anthropogenic N applied to watersheds enters rivers, their estuaries, and continental shelves, resulting in the production of N2 O (Fig. 1). Our analysis indicates that most emissions of N2 O take place in the Northern Hemisphere (63%), with largest contributions from the mid-latitudes; 37% of the global emissions are from the Southern Hemisphere. This is consistent with Prinn et al. (1990) who concluded from analysis of atmospheric trends between 1978 and 1988 that emissions into the 90°±30°N, 30°±0°N, 0°±30°S and 30°±90°S latitudinal belts contribute 22±34%, 32±39%, 20±29%, and 11±15%, respectively, to the global emis-
sions. Bouwman et al. (1995) also estimated, for their case 4b, that 60% of the total emissions are from the Northern Hemisphere and 40% from the Southern. Adding the emissions from rivers, estuaries and continental shelves to the Bouwman et al. inventory slightly improved the comparison to the Prinn et al., mid-point estimate for the North (63%)±South (37%) distribution as reported in Bouwman et al. (1995). Our estimated total global source (15 Tg N yÿ1 ) is higher than that of Prinn et al. (1990) (13 Tg N yÿ1 ), which is also consistent with the fact that Prinn et al. (1990) used a larger atmospheric lifetime for N2 O (about 170 year) than was used in more recent studies (120 years; Prather et al., 1995). As a result, Prinn et al. (1990) may have underestimated total global emissions.
278
S.P. Seitzinger et al. / Chemosphere ± Global Change Science 2 (2000) 267±279
N2 O emissions linked to N inputs from terrestrial sources versus N inputs from aquatic sources: A ®nal perspective from which to examine biogenic N2 O emissions is to consider whether the N driving the N2 O emissions in an ecosystem was originally from land or from aquatic sources. Aquatic N sources include biological N2 -®xation in aquatic ecosystems. In the framework of the current analysis, N2 O emissions associated with aquatic N sources include all ocean N2 O emissions (3.5 Tg N yÿ1 ) and that portion (55%) of continental shelf emissions that are associated with onwelling or advection of ocean N (0.3 Tg N yÿ1 ; Fig. 6) (Table 3). N2 O emissions attributed to a terrestrial N source include all biogenic N2 O from terrestrial ecosystems (2.5 Tg N yÿ1 anthropogenic, 6.6 Tg N yÿ1 natural soils), rivers (1.1 Tg N yÿ1 ) and estuaries (0.22 Tg N yÿ1 ) plus that portion of continental shelf emissions associated with atmospheric N deposition and river export (0.3 Tg N yÿ1 ) (Tables 1±3; Fig. 6). Thus, of the 14.5 Tg of biogenic N2 O±N emitted per year, approximately 75% (10.7 Tg N yÿ1 ) is due to terrestrial N and 25% (3.8 Tg N yÿ1 ) is from oceanic N sources.
4. Concluding remarks
Acknowledgements We would like to thank L. Bouwman and J. Olivier for assistance with obtaining the terrestrial and ocean data bases from EDGAR. This work was supported in part with funding from NOAA NJ Sea Grant (SPS) (NJSG-99-423), NOAA Coastal Ocean Program (SPS) and Wageningen University (CK). References
It should be kept in mind that the above analyses are based on N2 O emissions inventories currently available in gridded format. As noted above, there are a number of missing sources (Mosier et al., 1998), as well as more recent estimates of abiotic sources (Prather et al., 1995). Emissions from all sources have considerable uncertainty associated with them. As we improve our understanding of factors controlling N2 O emissions in ecosystems worldwide, most certainly our estimates will need to be reassessed. As anthropogenic activities associated with the use and cycling of N in the environment increase in the fu-
Table 3 N2 O emissions (Tg N yÿ1 ) linked to N inputs from terrestrial sources versus N inputs from aquatic sourcesa
a
ture, it will be particularly important to understand factors controlling N2 O emissions so that eective strategies can be developed to control emissions. The relatively large magnitude of freshwater and coastal marine N2 O emissions in combination with their large anthropogenic component indicates the need to obtain better quanti®cation of these sources and of the factors controlling them. This is illustrated by recent estimates of future N2 O emissions from rivers, estuaries and continental shelves which indicate that emissions from these sources alone may increase from 2 Tg N yÿ1 in 1990 to 5 Tg N yÿ1 by 2050 due to projected increases in N inputs to these aquatic ecosystems from anthropogenic activities (Kroeze and Seitzinger, 1998).
Region
N2 O from aquatic N
Oceans Continental shelves Estuaries Rivers Terrestrial soils Natural and anthropogenic Total
3.5 0.3
N2 O from terrestrial N 0.3 0.2 1.1 9.1
3.8
10.7
Inventories from Nevison et al. (1995), Bouwman et al. (1995) and Seitzinger and Kroeze (1998) (see text for details).
Bange, H.W., Rapsomanikis, S., Andreae, M.O., 1996. Nitrous oxide in coastal waters. Global Biogeochem. Cycles 10 (1), 197±207. Bouwman, A.F., Fung, I., Matthews, E., John, J., 1993. Global analysis of the potential for N2 O production in natural soils. Global Biogeochem. Cycles 7, 557±597. Bouwman, A.F., Hoek, K.W., Olivier, J.G.J., 1995. Uncertainties in the global sources distribution of nitrous oxide. J. Geophys. Res. 100, 2785±2800. Christensen, J.P., Murray, J.W., Devol, A.H., Codispoti, L.A., 1987. Denitri®cation in continental shelf sediments has major impact on the oceanic nitrogen budget. Global Biogeochem. Cycles 1, 97±116. Dentener, F.J., Crutzen, P.J., 1994. A three-dimensional model of the global ammonia cycle. J. Atmos. Chem. 19, 331±369. Edwards, M.H., 1986. Digital Image Processing of Local and Global Bathymetric Data. Department of Earth and Planetary Sciences, Washington University, USA. Erickson, D.J., 1993. Stability dependent theory for air±sea gas exchange. J. Geophys. Res. 98, 8271±8288. Food and Agriculture Organization Yearbook. Fertilizer, 1990, FAO Statistics Series No. 100, Vol. 40. Galloway, J.N., Schlesinger, W.H., Levy II, V., Michaels, A., Schnoor, J.L., 1995. Nitrogen ®xation: anthropogenic enhancement ± environmental response. Global Biogeochem. Cycles 9 (2), 235±252. Howarth, R.W., Billen, G., Swaney, D., Townsend, A., Jaworski, N., Downing, J.A., Elmgren, R., Caraco, N., Lajtha, K., 1996. Regional nitrogen budgets and riverine N & P ¯uxes for the drainages to the North Atlantic Ocean: natural and human in¯uences. Biogeochemistry 35, 75±139.
S.P. Seitzinger et al. / Chemosphere ± Global Change Science 2 (2000) 267±279 Jensen, H.B., Jorgensen, K.S., Sorensen, J., 1984. Diurnal variation of nitrogen cycling in coastal, marine sediments. II. Nitrous oxide emission. Marine Biology 83, 177±183. Khalil, M.A.K., Rasmussen, R.A., 1992. The global sources of nitrous oxide. J. Geophys. Res. 97 (D13), 14651±14660. Kroeze, C., Seitzinger, S.P., 1998. Nitrogen inputs to rivers, estuaries and continental shelves and related nitrous oxide emissions in 1990 and 2050: a global model. Nutrient Cycling Agroecosyst. 52, 195±212. Kroeze, C., Mosier, A., Bouwman, L., 1999. Closing the global N2 O budget: a retrospective analysis 1500±1994. Global Biogeochem. Cycles 13 (1), 1±8. Lerner, J., Matthews, E., Fung, I., 1998. Methane emission from animals: a global high resolution database. Global Biogeochem. Cycles 3, 281±285. Matthews, E., 1994. Nitrogenous fertilizers: global distribution of consumption and associated emissions of nitrous oxide and ammonia. Global Biogeochem. Cycles 8 (4), 411±439. Meybeck, M., 1982. Carbon, nitrogen, and phosphorus transport by world rivers. Amer. J. Sci. 282, 401±450. Mosier, A., Kroeze, C., Nevison, C., Oenema, O., Seitzinger, S., Cleemput, O., 1998. Closing the global atmospheric N2 O budget: nitrous oxide emissions through the agricultural nitrogen cycle. Nutrient Cycling Agroecosyst. 52, 225±248. Nevison, C.D., 1994. A model analysis of the spatial distribution and temporal trends of nitrous oxide sources and sinks. Ph.D. Dissertation, Stanford University, 218 pp. Nevison, C.D., Weiss, R.F., Erickson, D.J., 1995. Global oceanic emissions of nitrous oxide. J. Geophys. Res. 100, 15809±15820. Nishio, T., Koike, I., Hatorri, A., 1983. Estimates of denitri®cation and nitri®cation in coastal and estuarine sediments. Appl. Environ. Microbiol. 45, 444±450. Nixon, S.W., Ammerman, J., Atkinson, L., Berounsky, V., Billen, G., Boicourt, W., Boynton, W., Church, T., DiToro, D., Elmgren, R., Garber, J., Giblin, A., Owens, R.J., Pilson, M.E.Q., Seitzinger, S., 1996. The fate of nitrogen and phosphorus at the land-sea margin of the North Atlantic Ocean. Biogeochemistry 35, 141±180. Olivier, J.G.J., Bouwman, A.F., van der Maas, C.W.M., Berdowski, J.J.M., Veldt, C., Bloos, J.P.J., Visschedijk, A.J.H., Zandveld, P.Y.J., Haverlag, J.L., 1996. Description of EDGAR Version 2.0: A set of global emission inventories of greenhouse gases and ozone-depleting substances for all anthropogenic and most natural sources on aa per
279
country basis and on 1x1 degree grid. RIVM Report no. 771060002/TNO-MEP Report no R96/119, National Institute for Public Health and the Environment, Bilthoven, The Netherlands. Pilson, M.E., Seitzinger, S.P., 1996. Areas of shallow water in the North Atlantic. Biogeochemistry 35, 260±264. Pilson, M.E.Q., 1998. An Introduction to the Chemistry of the Sea. Prentice-Hall, New York. Prather, M., Derwent, R., Ehhalt, D., Fraser, P., Sanhueza, E., Zhou, X., 1995. Other trace gases and atmospheric chemistry. In: Houghton, J.T., Meira Filho, L.G., Bruce, J., Lee, H., Callander, B.A., Haites, E., Harris, N., Maskell, K. (Eds.). Climate Change 1994: Radiative Forcing of Climate Change and an Evaluation of the IPCC IS92 Emission Scenarios, the Intergovernmental Panel on Clmiate Change, Cambridge University Press, pp. 77±126. Prinn, R., Cunnold, D., Rasmussen, R., Simmonds, P., Aleya, F., Crawford, A., Fraser, P., Rosen, R., 1990. Atmospheric emissions and trends of nitrous oxide deduced from 10 years of ALE-GAGE data. J. Geophys. Res. 95, 18369± 18385. Seitzinger, S.P., 1988. Denitri®cation in freshwater and coastal marine ecosystems: ecological and geochemical signi®cance. Limnol. Oceanogr. 33 (4), 702±724. Seitzinger, S.P., Giblin, A.E., 1996. Estimating denitri®cation in North Atlantic Continental Shelf sediments. Biogeochemistry 35, 235±260. Seitzinger, S.P., Kroeze, C., 1998. Global distribution of nitrous oxide production and N inputs in freshwater and coastal marine ecosystems. Global Biogeochem. Cycles 12, 93±113. Van Aardenne, J.A., Kroeze, C., Hordijk, L., 1998. Analysis of the uncertainties in the IPCC default method for estimating N2O emissions from agricultural soils. In: Chan, K., Tarantola, S., Campolongo, F. (Eds.). Proceedings of SAMO, Second International Symposium on Sensitivity Analysis of Model Output, ISIS-SMA, EC Joint Research Centre, Ispra, Italy. Sybil SeitzingerÕs current research includes studies of: denitri®cation in rivers, estuaries and continental shelves; dissolved organic nitrogen inputs and bioavailability in aquatic ecosystems; and global modeling of N transport by world rivers and associated nitrous oxide production. Anthropogenic impacts are key components of her studies. She also enjoys windsur®ng, ice climbing, snow boarding, white water kayaking, and traveling.